Debayering a DSLR's Bayer matrix.

[RAC]

I had a play today and found the limits of the sensor thickness for testing sake. Its only in the corner and it still works. Looking at the flat photos i'm not holding my breath for how good this will be.

The best way to strip the sensor would be to have a small polishing mill that goes to a set depth. I had a play with a few old webcams(logitech 9000) and they have a sensor that is alot harder, after you pass the bayer filter there is nothing else to polish through, it stays flat and shiny with even colour unlike the Canon sensor. It would be interesting to play with other camers so i'll keep an eye out of dead ones for sale.

I'm hopping tomorrow night will be clear.

Mono 1000D by meg rac, on Flickr

[RAC]

I'm also interetsed to see how you've fitted that DSLR in the cooling box as it's something I might do with the 1000D I've got.

That camera is totaly stripped, its got a heater on the AA filter in front of the sensor and it doesn't frost over even in open air at -4degC. I'll take a shot of it next time i open the box.

[Spikey]

That camera is totaly stripped, its got a heater on the AA filter in front of the sensor and it doesn't frost over even in open air at -4degC. I'll take a shot of it next time i open the box.

Thanks, that would be really useful

[Gina]

I'm hoping I shall be able to see how the job is going as I do it without continuously putting the sensor back in a camera and connecting everything up. I may initially play with an old webcam which can be run with the sensor exposed - just take the works out of the case and remove the lens assembly. Much smaller sensor chip though.

[RAC]

I'm hoping I shall be able to see how the job is going as I do it without continuously putting the sensor back in a camera and connecting everything up. I may initially play with an old webcam which can be run with the sensor exposed - just take the works out of the case and remove the lens assembly. Much smaller sensor chip though.

One of the webcams i did never worked again as there were wires so small i could see them lol. I wish canon would make a mono dslr or you could buy the same sensors in colour or mono and just fit the mono one.

[Gina]

I have a whole host of old webcams that I don't want but if the sensors are that different from the Canon ones I'm not sure if testing on a webcam would be any use.

I've ordered the ScratchX 2.0 from Amazon plus some paint stripper I might try :- http://www.amazon.co...ils_o00_s00_i00

Both due to arrive tomorrow.

[martin_h]

Been looking at the DSS manual - this page :- http://deepskystacke...al.htm#rawdecod

No mention of not debayering but there are different ways of using the pixels. I've been using the "Super-pixel" mode with NB filters and I imagine this would be alright with the Bayer mask removed - giving a vast increase in sensitivity at the expense of resolution. The "Bayer drizzle" mode looks interesting though I don't quite understand what they're saying. This may give the required result with full resolution - not sure. Something to try though. Think I'll post in the Software forum - someone may know the answer.

I use DSS for stacking mono images from a CCD and just untick the fits box for the matrix....dunno if this would work for a converted colour camera..

[Gina]

I use DSS for stacking mono images from a CCD and just untick the fits box for the matrix....dunno if this would work for a converted colour camera..

Thank you - I'll try that

[Sp@ce_d]

Nebulosity's debayer routine is separate as well..

[Gina]

Nebulosity's debayer routine is separate as well..

Ah yes

[nytecam]

Hi RAC - interesting topic. As you've got two 1000Ds [one now debayered!] can you take identical shots - one through each with same exposure and share here with colour version greyscaled or better zero colour saturation. I'm concerned that loss of vital microlensing will lower sensitivity ! Apart from chemically stripping off the bayer colour layer [and microlensing too !] - any tests done with UV or other 'bleaching light' or vapour tested on a webcam etc that leave the microlensing untouched ? Just a thought ;-)

Edited October 29, 2012 by nytecam

[RAC]

A quick little test. The moons full so its not the best time for it but anyway.

Both these have 100% the same setting and are both single 2min subs. They were unquided and shot with my 400mm lens. Its a bit of the small magellanic cloud. The big change would come when doing any narrowband imaging with it.

If you click on the picks and go to my flickr page and click on the photo again you can swap between them with the arrow key.

Colour camera.

Color camera test by meg rac, on Flickr

Mono camera.

Mono camera test by meg rac, on Flickr

I'm having problems getting DSS to use the flats correctly.

[Gina]

Yes... The mono is just a bit lighter but not as much as your partly Bayer removed image would indicate. It will be more interesting to compare NB shots. My buffing paste is "Out for Delivey" so I should be able to make a start mysel today - I'm very much looking forward to that As you probably know I love extreme modding

[RAC]

I would recomend using something with a bit of surface area and try to get it as even as possible because i just cant get the flats to work as they should just yet. The more even you wear down the sensor the less you will need the flats. I'm sure i could make a small mill setup.

[Cath]

Very good project!

I'm just looking into the raw image format as I was going to write a little program to load and save the raw images without the debayering applied. I don't suppose you have the raw CR2 files of your images available do you RAC?

What I didn't know (until looking into this) is that the colour images produced from the raw data are interpolated. So when they say the camera is 15 mega pixels, it's the raw pixels they are refering too, and so the debayered colour result is really only 3.75 mega-pixels (and not 15 mpix) after interpolation. Tut tut.

So removing the CFA (the bayer layer) will increase your picture resolution to it's full 15 mpix (for a 15-mpix camera) grey scale, not that we can ever attain that kind of resolution here in the UK with our skies.

[Cath]

Do you think isopropanol could remove the colour layer without harming the micro-lenses?

Grinding away the colour layer seems a bit extreme to me

Edited October 30, 2012 by Cath

[Gina]

I could try that - I have a bottle of IPA, I use it for cleaning filters, sensor outer glass and for removing thermal paste etc. before applying new. I have two spare 1100D sensors so a bit of experimenting would be possible - but not too much as I don't want two dead sensors - that would leave only one's in use to have a go at. I can't afford too many dead cameras Though at the present rate of progress I might save myself the cost of an astro CCD camera.

[Cath]

Well don't go damaging one of your cameras Gina! .. I was just wondering if IPA would work.

Really need one of these for working at this scale ..

http://www.ebay.co.uk/itm/Dual-Stereo-Microscope-40x-80x-magnification-Dental-Labor-Diamont-Loupes-Loupe-/190712035030?pt=UK_BOI_Medical_Lab_Equipment_Lab_Equipment_ET&hash=item2c6752b6d6

I use one for SMT PCB's. They make ALL the difference.

[RAC]

Colour

https://www.dropbox.com/s/673uqm2xytsvett/IMG_0027.CR2

Mono

https://www.dropbox.com/s/ayfhwfm9f2jrzt3/IMG_0028.CR2

[Cath]

Colour

https://www.dropbox....tt/IMG_0027.CR2

Mono

https://www.dropbox....t3/IMG_0028.CR2

They are not complete RAC, corrupted.

[Cath]

oops, hold on.

Yes it's OK, sorry RAC

Edited October 30, 2012 by Cath

[Cath]

Quite a difference when you zoom onto a boundry where you've rubbed away the bayer layer.

[Gina]

Very interesting My gunge has arrived and I've found the sensor assembly that came out of my very first 1100D that's now all in bits So I'll soon be examining the sensor and taking it apart. I shall have to decide which working camera I'm going to put it in to see the result I guess the easiest would be the currently untouched one - my spare

Maximum risk £200 replacement value :- http://www.ebay.co.uk/itm/320885602263?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

[pwalsh61]

I usually drink my IPA

Oh, sorry, though you were referring to Indian pale Ale!

Interesting extreme mod. Not something I'm ever likely to try though!

[JamesF]

I have seen reports about it not being possible to dissolve CFAs with acetone, so I don't imagine IPA will work as a solvent. Mind you, I've also seen people claiming that it just isn't possible to do at all

James